Joint admission control and resource allocation in virtualized servers

In service oriented architectures, Quality of Service (QoS) is a key issue. Service requestors evaluate QoS at run time to address their service invocation to the most suitable provider. Thus, QoS has a direct impact on the providers' revenues. However, QoS requirements are difficult to satisfy because of the high variability of Internet workloads. This paper presents a self-managing technique that jointly addresses the resource allocation and admission control optimization problems in virtualized servers. Resource allocation and admission control represent key components of an autonomic infrastructure and are responsible for the fulfillment of service level agreements. Our solution is designed taking into account the provider's revenues, the cost of resource utilization, and customers' QoS requirements, specified in terms of the response time of individual requests. The effectiveness of our joint resource allocation and admission control solution, compared to top performing state-of-the-art techniques, is evaluated using synthetic as well as realistic workloads, for a number of different scenarios of interest. Results show that our solution can satisfy QoS constraints while still yielding a significant gain in terms of profits for the provider, especially under high workload conditions, if compared to the alternative methods. Moreover, it is robust to service time variance, resource usage cost, and workload mispredictions. (C) 2009 Elsevier Inc. All rights reserved.